Recap from OST LIVE with Trent McConaghy, Founder of Ocean Protocol — AI and Data Services on the Blockchain
With over 25 years of experience working in AI, Trent McConaghy started his career by doing AI research for national defense in the 1990s. He founded ADA, a startup that explored the relationship between human creativity and AI-based machine creativity. His latest startup is all about democratizing data for the planet. He then worked on ascribe (2013) and BigchainDB (2015). Now his focus lies on Ocean Protocol (2017), working on a shared public database for the internet.
Trent defines AI as an umbrella of different technologies that are still mysteries, yet to be converted into a repeatable discipline. Some of those umbrellas of technologies are things like fuzzy logic, form intelligence, artificial general intelligence, and genetic programming, in computer programs are automatically evolved to solve a variety of problems.
Relationship between AI and Data
In the early 2000’s, researchers discovered that adding massive amounts of data to an algorithm reduces the amount of error. Trent provides an example from his early work in AI. His work for the government of Canada had 20,000 data points, and he tested a variety of different algorithms to try to get better classification accuracy. He attempted neural networks, support vector machines, linear models, and genetic programming on a fixed data set and after several months of effort, he could only improve the accuracy from 55% to 65%. This means that there was still a 45% error making it non-deployable. Adding 10 to 100,000 times more data to the same algorithm reduced errors down from 40% to 30%, 20%, 10%, 5%, and even less than 1%. Google published a paper in 2009 called “The Unreasonable Effectiveness of Data” that showed several examples of this. AI that involves modeling such as deep learning, can really benefit from data.
The Data Problem
Trent says the demand side for data is insatiable and the big challenge lies on the supply side to make more data accessible. He mentions that there are a lot of enterprises sitting on petabytes of data and want to learn from it but are having trouble hiring researchers and data scientists. Why? Trent says it’s because AI researchers and data scientists know they’re valuable, so they’re off doing their own startups. Unfortunately, these AI researchers are too focused on algorithms that they forget they need data. It’s been very hard for them to monetize their AI startups. Trent says Google and Facebook and a small handful of others are exceptions to this problem because they’ve figured out how to monetize their data. They use AI as the refinery to extract value from the data. Trent says the challenge is that fortune 500 companies are trying to figure out what to do with their data and AI startups are crying to have it.
Ocean has a mission to create equalized access to data and AI services for the world. Trent believes there is a need for data marketplaces and data exchanges to emerge and compares a data marketplace to an airline ticket database. If you go to kayak.com and search for airline flights, you’ll see flight information appear. If you go to expedia.com, you’ll see the exact same flights appear, but it will be a different user interface. If you go to other travel sites, you’ll see a similar trend. Although the user interface might be different, it’s showing the same underlying data. This is because there’s one underlying core database that has this list of the airline flights that all the different airlines contributed to and it’s made available to the various last mile marketplaces like Kayak.
Ocean also incentivises for data commons. Trent believes that a lot of data out there is meant to be part of the commons, free to use, not priced, and people are incentivised to contributed to the commons. Ocean uses block rewards to help reward people to contribute data to the commons and then to verify that others have done so. The heart of Ocean is incentivizing towards the supply of data whether it’s priced or it’s free.
The Future of AI and Blockchain
Trent likes to paint models of what the future might look like in 20 years, 50 years, or 100 years from now. In 100 years, Trent believes there’s a very good chance that everything will be patterns of intelligence living on some substate. He believes that the AI is going to be there and hopes that the derivatives of human brains and minds will be there too. Trent mentions his worry with respect to jobs in the near term. In America, there are 3.5 million truck driver jobs and there are already production self-driving trucks on the road. Trent believes that in the next 5–10 years, autonomous trucks could decimate truck driver jobs, hotels, motels, and gas stations that truck drivers use. He also mentions the 6.5 million retail jobs that are under severe threat because of innovations like Amazon Go, the cashier-less Amazon store based in Seattle. Trent views these challenges as being in the near term.
Trent believes that Ocean can help to provide a transition, stating “if we can set things up where there’s a social safety net for the citizens as automated machines get set up, their surplus can be directed towards the social safety net that helps these people that have lost their jobs.” adding that in the near term “It’s really about helping a scientist solve better models, but as time goes on, it’s more about helping provide this benevolent substrate that helps humanity upgrade.”
Coming Up Next on OST LIVE: Pavel Bains
Be sure to join us next week for our OST LIVE with Pavel Bains, CEO of Bluzelle, a decentralized database that solves the security, integrity, reliability issues facing data management. Pavel will discuss the future of digital content and data on the blockchain. Subscribe to our YouTube channel or listen to the audio format on anywhere you listen to podcasts, including on iTunes, TuneIn, and Spotify. We’re also now available on Alexa! Simply add “OST LIVE” to your flash briefing.
OST blockchain infrastructure empowers new economies for mainstream businesses and emerging DApps. OST leads development of the OpenST Protocol, a framework for tokenizing businesses. In September 2018 OST introduced the OpenST Mosaic Protocol for running meta-blockchains to scale Ethereum applications to billions of users. OST KIT is a full-stack suite of developer tools, APIs and SDKs for managing blockchain economies. OST Ppartners reach more than 300 million end-users. OST has offices in Berlin, New York, Hong Kong, and Pune. OST is backed by leading institutional equity investors including Tencent, Greycroft, Vectr Ventures, 500 Startups.